Comment Re:Honesty (Score 1) 100
I have a repo in a sandbox with no remote. Claude is quite good at pushing commits to this repo during its own milestones during development, and it will use it to hunt down bugs when tests fail after a commit, stepping back to a commit before this session's work to see if this is a new bug , or a new way to expose an existing bug. If it sees that it is a new way to exercise an existing bug, it will binary search its way through the commits until it narrows down what changed, and then looks at how to fix it.
My overall strategy is very different. I don't think Claude's code is remotely good enough yet, so I don't trust it to go off and do "development" semi-unsupervised. I direct and review everything it writes, and I control the scope and content of every commit. To be a little more precise, Claude's code is generally fine (if a bit over-commented), but Claude rarely produces code structure that is clear, clean and correctly-mapped to the problem domain, to my (very picky) eyes. So I never just turn it loose on a large, multi-commit development task.